Ai · Top stories
Google Develops 'Frozen v2' AI Chip to Enhance Gemini Model Efficiency by 2028
Google is developing a new AI chip, called 'Frozen v2', designed to improve the efficiency of its Gemini models. Projected to be 6-10 times more efficient than current chips, it aims to address AI compute shortages and reduce dependency on external hardware. This development reflects a shift towards in-house chip production for AI workloads.
Google's AI Mode Integrates with YouTube Music, Canva, and Instacart for Task-Based Use
Google has expanded AI Mode's capabilities to interact with apps like YouTube Music, Canva, and Instacart. The updates allow users to manage tasks such as creating playlists, designing flyers, and compiling shopping lists directly within the app. This aims to enhance AI-driven task management and positions Google more competitively against other platforms offering similar app integrations.
Apple Intelligence Secures Chinese Market Entry with Alibaba's AI Integration
Apple has secured regulatory approval in China for its generative AI service, Apple Intelligence, by integrating local AI models from Alibaba and Baidu. The collaboration facilitates compliance with Chinese laws and advances Apple's expansion in a major market. This has implications for Apple's presence in China, a key growth region.
Inkling: Thinking Machines Releases Open-Weights Multimodal AI Model
Thinking Machines Lab has released Inkling, a multimodal AI model with approximately 1 trillion parameters. This open-weight model supports text, audio, and images, and features a mixture-of-experts design for efficiency. Its compatibility with a variety of inputs and adaptability through customization has positioned it as a flexible solution for enterprises and developers.
Alibaba Bans Anthropic's AI Tools Amid Espionage Allegations
Alibaba has banned Anthropic's AI tools, including Claude Code, for employees. The ban, effective July 10, follows Anthropic's accusation of a distillation attempt by Alibaba. This highlights rising tensions between U.S. and Chinese tech industries over AI technology and access.
Google Launches Faster AI Models for Image and Video: Nano Banana 2 Lite and Gemini Omni Flash
Google announced the launch of Nano Banana 2 Lite and Gemini Omni Flash, models designed to enhance multimedia processing efficiency. Nano Banana 2 Lite focuses on low-cost, fast image generation, while Omni Flash offers advanced video generation and editing. Available across Google platforms, these models aim to streamline creative workflows for developers and businesses using AI-generated content.
Netflix Launches AI Discover Tab for Mood-Based Content Recommendations in the US
Netflix has introduced a new "AI Discover" tab in its TV app in the U.S., offering users content recommendations based on their mood, viewing preferences, and available time. This feature aims to shift content discovery from endless scrolling to a more personalized, AI-powered experience, with themes like "something light" or "scare me." The AI Discover tab is a testbed for new features, powered by a new intelligence layer that responds to user input in real time.
Pangram Expands AI Detection Tools to Include Video and Image Content
Pangram, known for its AI text detection, has released new tools for identifying AI-generated images and videos. These updates allow subscribers to assess if video clips are partially or entirely AI-generated and automatically scan images for AI content via a Chrome extension. The company reports high accuracy rates in internal testing for both video and image detection.
Gemini Omni 1.2 Flash Released with Longer Video Creation and Price Reduction
Google has made Gemini Omni 1.2 Flash generally available, introducing native 15-second video creation, multi-turn conversational editing, and improved reference following. This version is 30% cheaper than Gemini Omni 1.1, costing $0.07 per second for 720p video, and is available across the Gemini API and Gemini for business.
Hybrid AI Model Combines Jev and Opus for Cost-Effective Identity Resolution
A new hybrid AI model combining Jev and Opus achieves high accuracy in identity resolution tasks at a significantly reduced cost. This approach leverages Jev's calibrated confidence to selectively query Opus, providing frontier model performance while being over four times cheaper than using Opus alone.
DOJ Establishes Super Intelligence Task Force to Combat AI Manipulation
The Department of Justice (DOJ) launched the Super Intelligence Task Force (DOJ-SITF) to prevent foreign and domestic actors from manipulating US-based artificial intelligence technology. This unit will address fraud and other malicious activities within the AI sector.
ToyBlocks: A New Open Simulation Framework for Robot Learning and Evaluation
A new open simulation framework called ToyBlocks has been introduced to provide a controlled environment for robot learning and evaluation. This framework allows researchers to generate tasks and training demonstrations automatically, enabling systematic study of robot behavior and failure causes.
BigQuery DataFrames Introduces .bigquery Accessor for Zero-Shot ML Predictions in Pandas
BigQuery DataFrames has released a new .bigquery accessor, allowing users to run BigQuery's AI functions directly within pandas DataFrames. This integration enables zero-shot machine learning predictions, such as forecasting and classification, by translating pandas operations into Google Standard SQL and executing them on BigQuery's infrastructure.
Canva Launches Craft, Its First Proprietary AI Image Generation Model
Canva has released Canva Craft, its first proprietary image generation model, developed by its research team. This model uses a two-stage process to interpret prompts and generate images up to 2K resolution in over 30 aspect ratios, aiming for precise object placement and legible text. Canva Craft was trained on human-designed templates from Canva's library, allowing generated images to be converted into layered designs using Canva Magic.
Sam Altman States Data Centers Offer Better RAM Utilization Than Personal PCs for AI
OpenAI CEO Sam Altman stated that having DRAM in personal computers for local AI inference is inefficient due to low utilization rates. He argued that centralizing RAM in data centers allows for maximum utilization, which is critical in a supply-constrained environment. This perspective supports OpenAI's cloud-based model business and suggests a preference for data center resource allocation over individual hardware ownership.
NVIDIA Introduces 64GB DGX Spark for Local AI Development
NVIDIA has released a 64GB unified memory configuration for its DGX Spark system, featuring the GB10 SoC. This new offering provides a more affordable and accessible option for local AI development, supporting models up to 100 billion parameters and addressing the reduced memory requirements of current efficient AI models. The 64GB DGX Spark models will be available from multiple manufacturing partners starting in October.
SpaceX Secures $40 Billion Debt Financing for AI Chip Purchases
SpaceX has closed $40 billion in debt financing to acquire GPUs for its artificial intelligence systems. This financing highlights the significant capital investment required for AI infrastructure, with chips expected to arrive in early 2027.
DeepSeek Releases V4.1 Flash 1017 Model with New CED Architecture and Asymmetric Parameter Activation
DeepSeek released V4.1 Flash 1017, a new sparse mixture-of-experts model built on its Causal Encoder-Decoder (CED) architecture. This model activates 8B parameters on input and 16B on output from a 552B-parameter backbone, and includes native image understanding, aiming for cost-efficient performance in agentic workloads.
Google BigQuery Introduces TimesFM 3.0 for Multivariate Time-Series Forecasting
Google has launched TimesFM 3.0 in preview for BigQuery, enabling native multivariate time-series forecasting directly within SQL. This update allows users to predict future values based on multiple related time series and external factors, addressing a limitation of previous univariate-only models.
IMF Chief Georgieva: AI is a key economic driver but also a growth threat
IMF Managing Director Kristalina Georgieva stated that AI is rapidly becoming a key driver of global economies, with potential to add half a percentage point to annual world growth. However, she warned that AI advancement, soaring energy costs, and record public debt are challenging global growth and increasing economic inequality.
McDonald's uses AI for dynamic pricing, varying burger costs by location
McDonald's is reportedly using AI to dynamically price menu items, adjusting costs based on estimated customer willingness to pay at each location. This system can lead to significant price differences for the same item between nearby stores, and franchisees report pressure to adopt the AI's recommendations.
Meta Open-Sources Muse AI Agent Code for Custom Hardware Development
Meta has open-sourced the code for its Muse AI agent, providing SDKs and firmware under the Apache 2.0 license. This allows users to integrate Muse with custom hardware projects using ESP32 boards or Raspberry Pi, enabling the creation of personalized AI gadgets and expanding Muse's application beyond Meta's devices.
Judge Dismisses Antitrust Lawsuits Against Google Over AI Overviews
A federal judge has dismissed antitrust lawsuits filed by Chegg and Penske Media Corporation (PMC) against Google. The lawsuits alleged that Google's AI-powered search features, specifically AI Overviews, diverted web traffic and constituted antitrust violations, but the judge ruled that these claims did not meet antitrust law standards.
Trump Administration Launches America.gov, an AI-Powered Government Website Using Gemini and Grok
The Trump administration launched America.gov, a new government website utilizing AI from Google's Gemini and Elon Musk's Grok. The site aims to improve user interaction with government services, addressing the 40 million daily interactions with existing government websites.
YouTube Expands AI Tools for Creators, Including Dynamic Thumbnails and AI Feedback
YouTube introduced new AI-powered features within its Studio app, including an expanded Ask Studio, AI feedback on video drafts, and dynamic thumbnail generation. These updates aim to help creators analyze content, optimize video performance, and grow their audience by providing suggestions on pacing, structure, storytelling, and content optimization.
Suno Releases v6 AI Music Models, Collaborating with Warner and BMG for Training Data
Suno has launched its new v6 family of AI music models, which were trained with assistance from Warner and BMG, marking the company's first formal partnership with music labels for model development. This release introduces new capabilities for AI music generation, including faster processing, advanced editing, and multi-modal understanding, impacting how users can create and manipulate AI-generated audio.
King Charles III Hosts AI Safety Summit with Tech Leaders and UK Officials
King Charles III convened a summit at Dumfries House in Scotland with leaders from Nvidia, OpenAI, and Anthropic, along with UK AI Minister Kanishka Narayan and other officials. The discussions focused on establishing shared principles for AI development, harnessing AI for societal benefit, and addressing the "existential dangers" of the technology if it falls into the wrong hands.
Anthropic Blocks Houthi-Linked Accounts Attempting AI-Assisted Missile Development
Anthropic identified and blocked accounts in Houthi-controlled Yemen that attempted to use its Claude AI model to develop advanced missiles, including a multi-variant hypersonic missile and a warhead with mobile phone hardware for guidance. The users conducted a failed guided rocket test and sought AI assistance to understand the failure, indicating attempts to advance their weapons capabilities.
New Mexico Lawyer Fined for Submitting AI-Generated Brief with Fabricated Testimony
The New Mexico Supreme Court fined attorney Stephen Aarons $5,000 and held him in contempt for submitting a legal brief containing fabricated police testimony and witnesses generated by OpenAI's ChatGPT. This incident highlights the risks of using AI tools in legal work without proper verification, as courts increasingly sanction lawyers for unvetted AI-generated content.
Google DeepMind Releases AlphaGenome Atlas for Human Genome Research
Google DeepMind has released AlphaGenome Atlas, an AI-powered tool that provides a predictive map of every possible DNA letter change in the human genome. This 1-petabyte dataset, generated by the AlphaGenome AI model, includes an AlphaGenome Variant Impact (AVI) score to help researchers understand how genetic mutations affect molecular biology and prioritize genetic research, particularly in non-coding DNA regions.
Cloudflare Introduces Tools for an "Agentic Internet" to Support AI Agents
Cloudflare has launched a suite of new tools and concepts, including Cloudflare Wallets, Kitesurf, WebMCP integration, Cloudflare Computer, and Precursor, to establish an "Agentic Internet." This initiative aims to provide AI agents with stable identities, native payment methods, specialized browsing capabilities, efficient runtimes, and enhanced security for interacting with web content and APIs.
Muse Spark 1.3 Released with Improved Agentic and Coding Performance
Muse has released Muse Spark 1.3, an update to its AI model, which offers enhanced performance in agentic workflows and coding tasks. This version is designed for better collaboration with users and more reliable execution of complex instructions, making it more practical for real-world applications.
Google Rolls Out AI Voice Features for Gmail, Docs, and Keep
Google has released new AI-powered voice capabilities, dubbed Docs Live, Gmail Live, and Keep Live, for its productivity suite. These features allow users to interact with their emails, documents, and notes using natural language queries and dictation, and are available to Google AI Plus, Pro, and Ultra subscribers on iOS and Android.
OpenAI faces 30 new lawsuits, including aiding and abetting claims, related to Tumbler Ridge shooting
Edelson PC filed 30 additional lawsuits against OpenAI on behalf of victims and families connected to the Tumbler Ridge mass shooting, bringing the total to 37. These new complaints, for the first time, accuse OpenAI of aiding and abetting the shooting, escalating the legal challenges beyond negligence claims. This development matters as it intensifies scrutiny on OpenAI's content moderation policies and its responsibility for user-generated content that may lead to real-world harm.
OpenAI Sees Executive Departures, Including COO Brad Lightcap and CRO Denise Dresser
OpenAI has experienced multiple executive departures, including longtime operating chief Brad Lightcap and Chief Revenue Officer Denise Dresser. Lightcap, who joined in 2018, left to "start something new," while Dresser, who joined in December, is being replaced by Dali Rajic. These changes occur as OpenAI prepares for a potential IPO and aims to justify its $852 billion valuation.
Google AI Mode Integrates Flight Price Tracking, Points/Miles Display, and Hotel Booking
Google has updated AI Mode in Search with new travel planning features, including flight price tracking, the ability to view flight and hotel costs in points or miles, and direct hotel booking capabilities. These additions allow users to manage travel logistics within the AI chat interface, expanding its utility beyond information retrieval.
Adobe Photoshop Introduces AI Assisted Editor and New AI Features
Adobe Photoshop is rolling out an update that includes an optional "AI Assisted Editor" interface, centralizing AI tools for image editing. New AI refinement capabilities, such as a "markup" feature for drawing changes directly on images and an enhanced "Instruct Edit with Masks" feature using the Firefly Image 5 model, are also being added. Additionally, Photoshop now includes Lightroom-style lighting controls, allowing for RAW-quality edits within the application.
Bill Gates Proposes 'Human Reserved' Jobs and AI Taxation Amidst AI Upheaval Concerns
Bill Gates, in an essay titled "The turbulent AI era is here. The choices we make are critical," proposed the concept of "human reserved" jobs to protect certain sectors from AI displacement. He also suggested taxing AI tokens and expressed concern that governments are unprepared for the broad impact of AI, which he believes will lead to one of the most turbulent times in human history.
Israel Creates Fake Think Tank to Influence AI Chatbots on Israel/Palestine Narratives
The Israeli Government Advertising Agency, through Piro, Inc., created the "Hanover Institute for Public Policy," a fabricated think tank producing reports designed to influence AI chatbots on topics related to Israel and Palestine. This initiative, termed "AI Story Optimization" by Piro and "LLM poisoning" by others, aims to shape the information large language models provide in response to user queries.
Google acquires Spirit Airlines operational data for AI model training
Google purchased a dataset from the defunct Spirit Airlines for $10 million to use in training its AI models. This acquisition provides Google with a large volume of real-world operational and business data, excluding personally identifiable information, to enhance its AI capabilities.